ParDRe

ParDRe identifies and removes duplicate and near-duplicate DNA sequencing reads from NGS datasets to reduce redundancy and improve downstream analyses such as variant calling, genome assembly, and transcriptome profiling.

Key Features:

• De Novo Parallel Processing: Operates in a de novo manner without relying on pre-existing sequence databases, enabling direct processing of raw sequencing data.
• Bitwise Suffix Comparison: Compares suffixes of DNA strings using a bitwise approach to detect duplicate and near-duplicate reads by focusing on terminal sequence characteristics.
• Hybrid MPI/Multithreading Architecture: Leverages a hybrid Message Passing Interface (MPI) and multithreading framework to distribute tasks across multicore systems.
• Performance Efficiency: Demonstrates up to 27.29× speed improvement over Fulcrum on platforms with two 8-core Sandy-Bridge processors, highlighting suitability for large-scale genomic projects.

Scientific Applications:

• Variant Calling: Reduces redundant reads to streamline variant calling workflows and decrease false positives from duplicated data.
• Genome Assembly: Eliminates duplicate reads to lower computational burden and improve assembly accuracy in large-scale sequencing projects.
• Transcriptome Profiling: Removes redundant RNA-Seq reads to facilitate quantification and differential expression analyses with reduced overhead.

Methodology:

Clusters sequences based on suffix similarity using a bitwise comparison technique in a de novo framework and employs hybrid MPI plus multithreading parallelization for scalable processing.